An integrated planar magnetic micropump

Abstract

This paper presents an integrated magnetic micropump that uses in-plane compliance-based check valves and a magnetically actuated membrane. The device, which allows for simple fabrication and system integration with other functional elements, consists of two functional layers both fabricated from poly(dimethylsiloxane) (PDMS). The upper PDMS layer provides a compliant membrane with an electroplated thin-film permalloy strip for actuation, while the lower PDMS layer incorporates microfluidic components including the microchannels, pump chamber, and a pair of check valves for flow regulation. The PDMS check valves, each having a compliant flap in contact with a stiff stopper to allow for unidirectional fluid flow with minimized leakage, are located at the inlet and outlet of the pump chamber, respectively. As such, the unidirectional flow at a controlled volumetric rate can be readily generated in accordance with the pumping actions. Systematic characterization of the micropump has been performed by studying the dependence of its pumping flow rate on the driving frequency of magnetic actuation, and the back pressure. Experimental results show that this micropump is capable of generating fluid flow of 0.15 μL/min at the frequency of 2 Hz, corresponding to a volume resolution of 1 nL per stroke, and working reliably against a maximum back-pressure of 550 Pa, demonstrating the potential application of this micropump for various integrated lab-on-a-chip systems.

1. Introduction

Lab-on-a-chip (LOC) microsystems integrate a wide variety of laboratory functions in a single miniaturized device, and offer advantages such as reduced sample consumption, improved system compactness and portability, and low cost. To date, LOC microsystems have been widely pursued for chemical analysis, environmental monitoring, molecular biology and medical investigations [1–4]. In such systems, biochemical solutions must be handled in minute volumes down to nanoliters, and such operations entail integrated micropumps for precise mobilization and control of fluid flow. To this end, a wide variety of micropumps using different fabrication techniques and actuation schemes have been developed [5].

The most commonly used micropumps for LOC systems are membrane-based [6,7] with actuation methods exploiting electrostatic [8,9], piezoelectric [10,11], thermopneumatic [12,13], shape memory alloy (SMA) [14], electromagnetic [15–18], and pneumatic [19,20] effects. In particular, magnetic micropumps hold great potential for fully integrated LOC microsytems with advantages of rapid time response, large displacement, and low actuation voltage. The earliest magnetic micropumps were created by silicon-based micromachining techniques, which are typically complicated and expensive [21]. Alternatively, polymeric materials, such as silicone [18], polycarbonate [22], poly(methyl methacrylate) (PMMA) [23], and poly(dimethylsiloxane) (PDMS) [24], have been used to fabricate magnetic micropumps. However, current polymeric magnetic micropumps mostly exploit hybrid designs involving multilayered structures with out-of-plane flow control elements, which still require complicated fabrication and packaging, and more importantly, difficulties in integration with other functional components. Recently, valveless magnetic micropumps based on nozzles and diffusers have been developed with simple planar design, fabrication, and integration [25–27]. These benefits, however, are accompanied by the lack of self-blocking due to the low diodicity (i.e. the ratio of forward flow rate to reverse flow rate) of the nozzle/diffuser components. Therefore, a back pressure at the outlet, typically existing in practice, may cause reverse fluid leakage, which is undesirable for LOC microsystems as the cross contamination between the upstream and downstream fluids may occur [5].

In contrast, integrated microfluidic systems using planar checkvalves for flow regulation [28–30] could potentially address these issues. We have previously used planar check valves to demonstrate a pneumatic micropump [31], although the use of pneumatic actuation significantly limited the device’s level of integration and portability. Here, we present an integrated micropump combining on-chip, membrane-based magnetic actuation with an electroplated permalloy thin-film strip, and in-plane compliance-based PDMS check valves for fluid regulation. The device is fully integrated in the sense that all microflow control components are integrated in a single layer, and the magnetic component is simply integrated within a thin membrane. As such, this device is miniaturized and allows for simple fabrication and easy integration with other functional elements. Moreover, the device is directly actuated through an externally applied magnetic field, providing an efficient and wireless operation method, desirable for applications to implantable biomedical systems. We have characterized the micropump by systematically studying the pumping flow rate at varying operational parameters such as the driving frequency, magnetic field strength, and back pressure. Experimental results demonstrate that the micropump is able to produce continuous flow with a volume resolution of approximately 1 nL at the frequency of 2 Hz.

2. Design and principle

The micropump mainly comprises two functional layers (Fig. 1a): a compliant PDMS membrane with a thin permalloy strip on top for magnetic actuation, and a lower PDMS functional layer incorporating flow control elements, including planar check valves, a square-shaped pump chamber (4 mm × 4 mm × 160 μm in length, width, and height, respectively), and flow channels (400 × 160 μm² in width and height). These two functional layers are then sandwiched between a flat PDMS or glass substrate and a PDMS encapsulation layer that is used to cover the PDMS membrane. The two check valves are placed, respectively, at the inlet and outlet of the pump chamber (called “inlet valve” and “outlet valve”) to regulate the flow in a unidirectional manner. The permalloy actuation strip has a dimension of 3000 × 1740 × 20 μm³ in length, width, and thickness, respectively, and covers one half of the square membrane (4 mm × 4 mm × 20 μm in length, width and thickness), with an edge-to-edge distance of 500 μm from the membrane edges (Fig. 1b).

Fig. 1.

Schematic of the magnetic micropump design: (a) multilayer structure, (b) layout of permalloy actuator (top view) on a 4 × 4 mm² membrane, and (c) detailed design of the check valve (top view).

The check valves in the micropump exploit a simple planar configuration [32] in which a compliant flap (60 × 350 × 160 μm³ in length, width, and height, Figure 1c) is as-fabricated in contact with a stiff stopper (250 × 340 μm² in length and width). Under forward pressures (from the inlet to the outlet as shown in Fig. 1), the flap is pushed away from the stopper, allowing fluid passage. While under reverse pressures (from the outlet to the inlet), the flap remains in firm contact with the stopper, shutting off the flow. Therefore, unidirectional flow can be readily achieved with this in-contact minimized leakage check valve.

Fig. 2 illustrates the operation of the magnetic micropump. It generally consists of two-mode cycles, i.e. priming mode and pumping mode. In priming mode, the device is placed in a magnetic field generated, for example, by an external electromagnet, the magnetic field then produces a magnetic torque on the magnetized permalloy strip, which is directed along the width of the permalloy, and causes the membrane to deflect upward, generating a negative pressure in the pump chamber to open the inlet valve for the fluid to be introduced into the pump chamber, while to shut off the outlet valve to prevent the reverse flow from outlet channel to the pump chamber (Fig. 2a). Next, in the pumping mode, when the external magnetic field is switched off or reversed, the membrane recovers or deflects downward, creating a positive pressure to push the fluid in the pump chamber to the outlet channel. In this case, the outlet valve is under a forward pressure and thus is opened for fluid flow, while the inlet valve is under a reverse pressure and thus checks the reverse flow back to the inlet (Fig. 2b). Hence, by applying a periodic magnetic field, this micropump is capable of producing unidirectional fluid flow continuously.

Fig. 2.

Actuation and operating principle of the magnetic micropump.

3. Experiment

3.1. Fabrication process

The fabrication of the micropump device started with the top encapsulation layer using the standard PDMS soft lithography technique [33], in which the SU-8 negative photoresist (SU-8 2100, MicroChem Corp.) was used to fabricate the molding master. Next, for the permalloy electroplated PDMS membrane (Fig. 3), a polyethylene (PE) sheet was first laminated to a 4-in. silicon wafer using double-sided tape. The mixed PDMS prepolymer and curing agent (Sylgard 184 Silicone Elastomer Kit, Dow Corning) at a weight ratio of 1:10 was then spin-coated onto the PE sheet at a speed of 2000 rpm and cured at 70 °C for about 1 h, obtaining an approximately 20-μm thick membrane (Fig. 3a). A 100-nm thick copper (Cu) seed layer was then thermal evaporated on the membrane, followed by spin-coating and patterning of a 20-μm thick Shipley Microposit S1818 photoresist layer (see the photolithography parameters in Table 1) to define the permalloy deposition area (Fig. 3b). The permalloy was then deposited by electroplating (Fig. 3c, see the electroplating conditions in Table 2). The S1818 photoresist was then stripped with acetone, during which the Cu seed layer served to prevent acetone solution from penetrating through the PDMS membrane. The resulted PDMS/Cu layers were detached from the handling wafer and transferred to Cu etchant solution to etch the Cu layer. After the unused Cu was removed, the remaining PE/PDMS/Cu/permalloy stack was irreversibly boned to the top PDMS encapsulation layer (Fig. 1a). Followed by peeling off the PE sheet, a thin permalloy integrated PDMS membrane with minimum distortion was finally obtained (Fig. 3d).

Fig. 3.

Fabrication process of the actuation PDMS membrane with electroplated permalloy strip: (a) spin-coating PDMS membrane on a transparency film; (b) depositing a Cu seed layer and pattering photoresist; (c) electroplating permalloy on the Cu seed layer; and (d) peeling off the PDMS membrane from the transparency film after the removal of remaining photoresist and seed layer.

Table 1.

The S1818 photolithography parameters.

Step No.	Description	Equipment	Conditions
1	Spin coat S1818	Laurell WS-650–23 Spin Coater	500 rpm for 30 s
2	Soft bake	Hotplate	90 °C for 30 min
3	Exposure	Suss Mask Aligner MJB3	60 s
4	Develop	Fume hood	MIF 300 until clear (~1 min)
5	Rinse in	Fume hood	DI-water for 1 min
6	Dry	Fume hood	Nitrogen flow until dry (~30 s)
7	Hard bake	Oven	90 °C for 50 min

Table 2.

Bath composition and operating conditions of electroplating.

Bath composition	NiSO4·6H2O	200 g/L
	FeSO4·7H2O	8 g/L
	NiCl2·6H2O	5 g/L
	H3BO3	25 g/L
	Saccharin	3 g/L
Electroplating conditions	Current density	40 mA/cm2
	Time	45 min
	pH	3.2
	Temperature	30 °C
	Agitation	Magnetic agitation

Subsequently, a particular through-opening PDMS replica method [34] was used to fabricate the lower PDMS functional layer including the pump chamber, check valves, and microchannels. After peeling off the resulting thin layer from the master, the compliant flaps in the check valves were directed to contact with and remain stuck to the stoppers by spontaneous interfacial adhesion or manual manipulation [31].

Next, to facilitate the alignment during device assembly, two small permalloy pillars with a diameter of 400 μm, were electroplated on the PDMS membrane, while two holes in the same size were patterned the lower PDMS functional layer. After the two layers were cleaned with acetone, they were placed into a shallow dish with methanol as a surfactant for alignment. The two layers were then aligned and assembled manually under a microscope. After the methanol evaporated, the two layers were reversibly bonded together. The accuracy of the alignment was estimated to be about 5–10 μm, by comparing the actual distance of the permalloy strip’s edge to the membrane (Fig. 1b). Then, following an oxygen plasma treatment on the interfacial surfaces, the four structural layers were boned together in the top–bottom sequence as shown in Fig. 1a. The micropump device was finally obtained by connecting the inlet and outlet with plastic tubings (Tygon). A packaged device with the dimension of about 1.5 × 1.5 × 0.4 cm³ is shown in Fig. 4.

Fig. 4.

(a) Image of a check valve with in-contact flap and stopper. (b) Image of a packaged magnetic micropump device.

3.2. Test setup

To test the micropump (Fig. 5), a permanent magnet bar (magnetic field strength 200 kA/m) was placed in parallel to the device surface to allow saturated magnetization of the permalloy strip. An electromagnet (RE-303012, Magnetic Products Inc.) was then placed 2 mm below the permalloy strip. By applying square-wave periodic current generated by a waveform generator (33220A, Agilent) to the electromagnet, alternating magnetic field was induced to drive the magnetized permalloy strip.

Fig. 5.

Schematic of the experimental setup for micropump characterization.

During experiments, de-ionized water was used as a sample fluid. To measure the pumping rate, a high-speed CCD camera (PLB742U, PixeLINK) was used to record the small movement of the pumping water meniscus in a tubing that was connected to the outlet. And then the flow rate was obtained by averaging the water traveling speed. To test the micropump against backpressures, a pressure difference was generated by placing the inlet tube and outlet tube at a specified height difference ΔH, and the pressure difference (i.e., backpressure) could be calculated as ΔP = ρgΔH, where ρ is the density of the de-ionized water and g is gravitational acceleration.

4. Results and discussion

This section presents the experimental results of characterizing the magnetic micropump device. We first characterized the integrated check valve, exhibiting its efficacy in minimizing the fluidic leakage and regulating the unidirectional flow. We then characterized the pumping flow rate of the micropump as functions of the frequency of the magnetic actuation, the applied current to the driving electromagnet, and the backpressure. The results demonstrate that this micropump is capable of precisely generating fluid flow continuously. During experiments, five device were tested and characterized with error bars (shown in the data figures) representing the standard deviation.

4.1. Check valve characterization

The integrated check valve was first characterized by studying the forward and reverse flow rates under static pressures applied over the check valve, ranging from 0 to 10 kPa (Fig. 6). It can be seen that the forward flow rate remained negligible until it reached a pressure of approximately 1 kPa, which can be considered as the critical pressure for opening the check valve. During the pressure range of 1–10 kPa, the forward flow rate exhibited a linear relation with the pressure, implying the forward flow resistance of the check valve (i.e., the ratio of applied pressure to flow rate) stays nearly constant within this pressure range. For the flow under reverse pressures, there was virtually no fluid flow detected at all pressures during the experiment. This minimized leakage is most likely attributed to its normally in-contact flap-stopper configuration, by which a zero-gap under reverse pressures is formed to maximally shut off fluid passage. As such, the check valve can efficiently regulate the fluid flow in a unidirectional manner.

Fig. 6.

Forward and reverse flow rates through the check valve under different applied pressures.

4.2. Micropump characterization

We further investigated the performance of the magnetic micropump under varying operational parameters, including the applied electrical current to the driving electromagnet and backpressure.

The frequency response of the pumping flow at varying applied currents was first studied (Fig. 7). It can be observed that the relationship between the pumping rate and driving frequency under different applied currents was consistent. In particular, the pumping rate initially increased with driving frequency in the lower range, and then reached its maximum at approximately 2 Hz, which can be determined as the optimal driving frequency. This initial increase of pumping rate with frequency can be explained by the fact that the membrane deflects more rapidly with the increasing driving frequency in the lower frequency range, resulting in an increased pumping rate due to the speeded membrane vibration. After 2 Hz, the pumping rate steadily decreased as the frequency further increased. This is because the deformation of the membrane could not sufficiently correspond to the external driving frequency at higher frequencies. The mismatch between the membrane deflection and driving frequency became even more significant at driving frequencies beyond the optimal driving frequency, and eventually attenuated the membrane deflection, causing a reduced stroke volume indicated by the gradual decrease in the pumping rate as the frequency increased above 2 Hz (Fig. 7).

Fig. 7.

Measured pumping flow rate as a function of the frequency at which an electric current of amplitude 0.15, 0.20 or 0.25 A was used to drive the external electromagnet.

Furthermore, we observed an increase of the pumping rate with the applied current to the driving electromagnet. Maximal low rate of 0.15 μL/min, 0.21 μL/min, and 0.26 μL/min were achieved at applied currents of 0.15 , 0.20 , and 0.25 A, corresponding to a power consumption of approximately 0.2 , 0.4 , and 0.6 W, respectively. This trend was due to that a larger applied current could induce greater driving magnetic field strength, and in turn, an increased membrane deflection for a greater pumping rate. Note that when the flow rate of 0.15 μL/min was achieved at the driving frequency of 2 Hz and applied current of 0.15 A, the mean volume per one stroke was calculated to be approximately 1 nL. This indicates that the magnetic micropump is capable of delivering fluid samples with a volume resolution of 1 nL, which is much smaller than those by existing devices [21,23,29]. Therefore, this micropump can be potentially employed for fluidic systems that require accurate delivery of minute volumes of samples in the areas such as genomics, proteomics and neurobiology [18].

We then investigated the performance of the micropump as a function of backpressure (Fig. 8). The experiment was conducted under a fixed applied current of 0.25 A and varying frequencies of 2 Hz, 3 Hz, and 4 Hz. It can be seen that the pumping rate decreased linearly with the backpressure at all driving frequencies, which is consistent with the linear relationship of the forward flow with the applied pressure for the check valves (Fig. 6). At zero pumping rates, a maximum backpressure of 550 Pa was obtained at a driving frequency of 2 Hz, while 490 Pa and 420 Pa for 3 Hz and 4 Hz, individually. The decrease of the maximum backpressure with driving frequency is most likely due to the reduced membrane stroke volume at frequencies beyond the optimal driving frequency as discussed above.

Fig. 8.

Measured pumping flow rate as a function of the back pressure at different driving frequencies.

Compared with our previous pneumatic pump [31], the mechanical characteristics (such as flow rate as a function of pressure, backpressure, and driving frequency) are very consistent, while the maximum pumping flow rate and back pressure are both significantly smaller: 0.25 μL/min and 0.55 kPa, compared with 41 μL/min and 25 kPa, respectively, by the previous pneumatic pump. This decrease is likely resulted from the fact that the magnetic actuation (a few hundred μN) is usually weaker than pneumatic actuation (a few mN to N).

Furthermore, the long-term reliability of the micropump was also characterized. Five devices were tested with 10,000 work cycles continuously at a current of 0.25 A and a driving frequency of 2 Hz. No failures were observed, and the valves and PDMS membranes were operated stably throughout. These results demonstrate that this integrated magnetic PDMS pump holds the potential for application of implantable MEMS drug delivery in genomics, proteomics and neurobiology.

5. Conclusion

This paper presents a miniaturized magnetic micropump for integrated fluid manipulation in LOC systems. The device consists of two in-plane, compliance-based check valves located at the inlet and outlet of the pump chamber, and a thin PDMS membrane electroplated with a permalloy thin-film strip on top for magnetic actuation. The device is fully integrated that all microflow components including the microchannels, pumping chamber, and check valves are incorporated into one single layer, while the magnetic actuation component is integrated within a thin membrane. This device is therefore simple in device fabrication and integration with other microfluidic components. The device is driven by a magnetic field remotely applied by an electromagnet and is amenable to implanted applications where it is desirable to eliminate the wiring between the device and power source.

The micropump has been systemically characterized using de-ionized water as a sample liquid, with the pumping flow rate determined by the average speed of the water meniscus in the outlet tubing. The check valves in the micropump device were first investigated, demonstrating their efficacy of regulating unidirectional flow with virtually no leakage detected. Furthermore, the micropump has also been characterized by studying its frequency-dependent flow rate at different applied currents to the driving electromagnet and varying backpressures. The experimental results have indicated this micropump was able to produce fluid flow at a volumetric resolution of approximately 1 nL per stroke, and operate reliably against a maximum backpressure of 550 Pa. These results demonstrate the potential of this device for use in LOC microsystems where integrated micropumps are desired.